Balint Gucsi MINDS CDT, 2022

Doctoral Researcher, University of Southampton

Balint Gucsi is a doctoral researcher working on human-robot interaction, and a specific problem within it: robots that behave the same way with everyone tend to frustrate people whose preferences, tolerances and communication styles fall outside the assumed norm.

His research develops frameworks that allow robots to behave in a user-aware way, adapting their responses during an interaction based on signals from the person they are working with. Those signals range from explicit feedback to more subtle multimodal cues, and the approaches Balint develops draw on multi-armed bandit learning, constrained optimisation, multimodal inference and research into social signal processing and the psychology of frustration.

The work spans a range of interaction types, from navigation scenarios to collaborative tasks, and has been tested in multiple user studies involving TIAGo and Pepper robots.

His publications include work presented at HRI 2025 in Melbourne, ICRA 2025 in Atlanta, RO-MAN 2025 in Eindhoven, and a paper accepted at ICRA 2026. Two multimodal datasets from this research are publicly available via Zenodo.

Balint is due to graduate in 2026.